| Renewable energy sources, such as solar and wind power, have been widely used in various forms of networks, which include both communication networks and the smart power grid. Since the renewable energy is usually highly fluctuating, the temporal and spatial dynamics of such sources are often difficult to model and predict. This makes it extremely challenging to incorporate and operate such energy sources while maintaining adequate utilization of them. In this thesis, we study the control and optimization strategies of renewable energy in several network scenarios, including wireless sensor networks, WiFi networks, and the smart grid.;We first investigate the problem of maximizing the throughput or utility performance for a wireless sensor network with renewable energy. The two solutions that we developed here are based on a finite time horizon and an infinite time horizon respectively. In the finite-horizon problem, we propose a simple heuristic distributed scheme in a rechargeable sensor network with optimal performance under some specific conditions and with provable performance guarantee. Regarding the infinite-horizon problem, after characterizing the optimal network utility with an upper bound, we develop a low-complexity online scheme that is asymptotically optimal as network operation period of time tends to infinite. It shows that by focusing on long-term system performance, we can greatly simplify computational complexity while maintaining high performance.;We then study the problem of efficient channel access in WiFi networks, where the devices are powered by heterogeneous energy supplies, such as battery, electric power, and/or renewable energy. We propose a lifetime adjustable design for such WiFi networks, where each client turns off its radio to save energy when it senses a busy channel, and sleeps for a random time period before sensing the channel again. It is proven that our scheme achieves near-optimal proportional-fair utility performance for single access point (AP) scenarios. Moreover, it also alleviates the near-far effect and hidden terminal problem in general multiple AP scenarios heuristically.;Beyond the usage in communication networks, renewable energy can be also utilized in the smart power grid. We focus on an end-user with renewable energy harvesting devices, when the market electricity price is time-varying. Two problems are studied under this framework. First, we propose a scheme to minimize the energy cost for which usage of electricity appliances are subject to their individual delay constraints and a long-term average delay constraint. In the second problem, we allow end-users to trade energy with the grid. Similar to the stock market, end-users can buy and store energy from the grid when the electricity price is low and sell the energy to the grid when the electricity price is high. We propose a scheme, which is shown to be asymptotically optimal.;Our investigation in these application scenarios of renewable energy yields different algorithms with provable performance. These results show the significance of exploiting new approaches to incorporate and manage renewable energy while maintaining their high utilization. |
